Utilizing CO2 as a strategy to scale up direct air capture may face fewer short-term barriers than directly storing CO2

Direct air capture (DAC) is increasingly recognized as a necessary puzzle piece to achieve the Paris climate targets. However, the current high cost and energy intensity of DAC act as a barrier. Short-term strategies for initial deployment, technology improvement, and cost reduction are needed to enable large-scale deployment. We assess and compare two near-term pathways leading to the same installed DAC capacity and thus yielding the same cost reductions: its combination with CO2 storage as direct air carbon capture and storage, or its deployment for CO2 utilization as direct air carbon capture and utilization e.g. for synthetic fuels, chemicals, and materials; we characterize these as Direct and Spillover pathways. Drawing on the Multi-level Perspective on Technological Transition as a heuristic, we examine both technical and immaterial factors needed to scale up DAC under the two pathways, in order to assess the pathways’ relative advantages and to identify possible short-term bottlenecks. We find neither pathway to be clearly better: the Direct pathway offers technical advantages but faces regulatory barriers that need to be resolved before deployment, while the Spillover pathway offers market and governance advantages but faces challenges related to hydrogen production and increasing resource needs as it scales up. There may be reasons for policymakers to therefore pursue both approaches in a dynamic manner. This could involve prioritizing the Spillover pathway in the short term due to possibly fewer short-term regulatory barriers and its ability to produce net-zero emission products for existing and accessible markets. Once short-term governance obstacles have been addressed, however, the Direct pathway may allow for more efficient scaling of DAC capacity and cost reductions, especially if by then the needed infrastructure and institutions are in place.


Introduction
To limit warming below 2 • C and especially 1.5 • C, carbon removal will likely be needed, first to counterbalance greenhouse gas emissions from hard-to-abate sectors such as aviation, steel, cement and chemical industry, and agriculture [1], and achieve net-zero emissions, and second to compensate for temporary temperature overshoot [2].Direct air carbon capture and storage (DACCS) will likely play an important role in the future carbon removal toolset [3][4][5][6][7] due to its large carbon removal potential [8] and its ability to sequester atmospheric CO 2 independently from point sources (e.g.bio-or fossil-fueled smokestacks) [9].DACCS is however currently expensive, typically costing $500-3000/tCO 2 removed [10,11], which is likely to be a major barrier to its deployment.
Around 80% of the current cost of DACCS arises from the direct air capture (DAC) process [10][11][12][13][14][15][16].Estimates of the learning rate (i.e.cost reduction for each doubling of cumulative installed capacity) expected for DAC technologies lie at 10%-15% [11], similar to wind power but lower than historically observed for PV [17,18].To stimulate learning effects and concomitantly lower costs, large-scale deployment of DAC technology is essential [15,18].This could be achieved by directly deploying DACCS for carbon removal, which would result in immediate negative emissions.However, for nearterm carbon removal needs, DACCS is not costeffective and will not be in the near future, compared to cheaper and more beneficial alternatives such as afforestation [19,27,28].Moreover, as long as emissions are high, reducing emissions is a cheaper abatement option than removing atmospheric carbon with DACCS [29].A new, separate market, or another policy approach altogether, is therefore required for large-scale deployment of DACCS [30,31].
One alternative strategy would be to support DAC deployment for CO 2 utilization (as direct air carbon capture and utilization, DACCU) to produce synthetic fuels and chemical intermediates.While this strategy does not immediately lead to negative emissions and requires hydrogen supply and fuel production capacities, it makes use of existing demand and markets for hydrocarbons (e.g.fuels, naphtha) [32], just as support for wind and solar power make use of existing demand for electricity.Supporting DACCUbased hydrocarbons would lead to the same effects for DAC, enabling learning for DACCS indirectly as a technology spillover effect [33,34].
Here, we aim to fill this gap in the literature by performing, for the first time, a holistic comparison of two applications for the initial development and scale-up of DAC: either as DACCS for carbon removal (the Direct pathway), or as DACCU for CO 2 utilization (the Spillover pathway).While both strategies ultimately result in DAC deployment and attain the same DAC capacity and maturity in the long run, they diverge significantly in the challenges they present in the short term.These challenges encompass not only the technical and systemic aspects but also extend to crucial considerations such as infrastructure development, institutional dynamics, and the intricate interplay of social and behavioral factors.This study goes beyond prior assessments that solely focused on technical and economic aspects by providing a comprehensive comparison of these pathways.Recognizing that real-world technology transitions are influenced by factors beyond pure engineering and economics aspects [69,70], our study provides a holistic and multidimensional evaluation and comparison of these two diffusion pathways across a vast range of pivotal socio-technical factors.By doing so, our research can significantly inform the decisions of policymakers, investors, and DAC suppliers and thus contribute to the acceleration of DAC deployment.

DAC development pathways
We consider two pathways for the early deployment of DAC (figure 1).In the Direct pathway, air-captured CO 2 is compressed, transported and stored in sedimentary basins or basaltic rocks, where it is trapped via in-situ mineralization.In the Spillover pathway, the captured atmospheric CO 2 is reduced to CO and then combined with green hydrogen to produce synthetic fuels (e-kerosene, e-diesel, e-gasoline) and chemical intermediates (naphtha) via Fischer-Tropsch synthesis.We restrict our CO 2 utilization scenario to synthetic fuels and chemical intermediates as these are likely to represent the largest markets for CO 2 utilization [46], and exclude other applications (e.g.food, urea).To capture the variability in the performance of the Spillover pathway due to different process technologies, we assess two different technologies for electrolysis (alkaline and polymer electrolyte membrane) and CO 2 reduction (electrochemical CO 2 reduction and reverse water-gas shift) (cf supplementary methods 1.1 for a detailed description of the technologies).
Although the pathways involve different technologies-particularly for fuel production (Spillover) or storage (Direct)-with different environmental impacts and fundamentally different end products, both pathways achieve the same goal of 1 GtCO 2 /year of DAC deployment, thus bringing DAC costs and maturity to the same point on the learning curve.We assume that the pathways have identical initial annual DAC capacities-0.3ktCO 2 in 2020 [71]-and then follow an exponential curve, reaching 67 MtCO 2 in 2030 on the pathway to 1 GtCO 2 in 2050.This assumption is based on the observation, common for many technologies, that technologies follow a S-shaped deployment [72]: Since 1 GtCO 2 is only about 20%-50% of what is likely to be needed by the end of the century [4], when the S-shaped deployment curve flattens, we assume that DAC deployment follows an exponential phase until 2050.We base the target DAC volume of 1 GtCO 2 by 2050 as a compromise between conservative estimates of the total expected market for CO 2 -based products by 2060 (0.9 GtCO 2 /year) [65] and optimistic estimates of over 3 GtCO 2 of CO 2 -demand for utilization that DAC will need to meet by 2050 [46].An installed DAC capacity of 1 GtCO 2 by 2050 is also roughly consistent with Integrated Assessment Model projections of DACCS deployment by 2060 to meet temperatures below 2 • C [73].To assess the relative advantages and disadvantages of the two pathways, we analyze the entire process from input CO 2 to conversion and outputs, excluding DAC itself, as it is a common element in both pathways.

Analytical framework
We take a sociotechnical systems approach to compare the DAC development pathways, building on the multi-level perspective on sociotechnical transitions [69,74,75], and emphasising not only the technologies but in particular the social embedding of technologies in a complex system that includes all factors, both material and immaterial, that affect and enable the technology [76].We assess the factors needed to scale up DAC under the two pathways, evaluating short-term bottlenecks and the relative advantages and disadvantages of each pathway.We review technical factors (technological maturity, resource use determined by technology efficiency, and transport infrastructure) and immaterial factors (markets, regulations, international governance, and social and behavioral factors).
The technologies are identical for the DAC component but differ for the rest of the process.While the Direct pathway requires the parallel deployment of CO 2 compression, transport, and storage technologies, the Spillover pathway requires electrolysis, CO 2 reduction, transport of key inputs, and Fischer-Tropsch synthesis.As a result, the factors that enable the production, trading, and governance of DAC-based carbon removal are fundamentally different from those that enable production, trade, and governance of DAC-based hydrocarbons (i.e.fuels and chemical intermediates).However, our analysis should be interpreted only in the context of the development of DAC, and not as an exercise in determining which technology should be deployed in absolute terms.Indeed, the two DAC deployment pathways lead to fundamentally different end products (carbon removals in one case, and hydrocarbons to replace fossil ones in the other), and they are not mutually exclusive.
For each factor, we formalize evaluation criteria adapting from the framework by Förster et al [77] and identify what short-term bottlenecks would impede the immediate deployment of DAC (table 1).Then, we review both the academic and grey literature for each sociotechnical element and assess them based on the evaluation criteria and our definition of bottlenecks.For a comprehensive review of the literature underpinning our Results, refer to supplementary methods 1.2 and supplementary results 2.2-2.5.Whenever possible, we evaluate factors both in the short-term (2030) and long-term (2050).Finally, we compare the two pathways to establish their relative advantages and disadvantages.
Among technical factors, we first assess technological maturity of the whole conversion route, particularly based on the technological readiness level (TRL) of the least mature key process technology (cf supplementary methods 1.2.1 for a detailed account of the TRL scale used).We identify a short-term bottleneck if a key technology has a TRL below 5 ('large prototype maturity' [78]).
For resource use, we quantify environmental impacts, namely net greenhouse gas emissions over the life cycle, water, and land use, and resource use, namely energy and capital investment needs for the process technologies of each pathway.We first model an exponential deployment from current installed DAC capacity to 1 GtCO 2 in 2050 and calculate net greenhouse gas emissions, resources and capital investment needs in each year, which are influenced by energy efficiency, grid decarbonization, and technological learning (cf supplementary methods 1.2.2).We consider exceeding 20% of the installed capacity of/investments in global renewable energy production in 2021 (8000 TWh [79] and €500 billion [80] in 2021) as the threshold for the energy consumption and capital investments in 2030 for the technologies directly involved in the pathways (cf figure 1).While this threshold was arbitrarily chosen, spending over 20% of current investments in renewable energy corresponds to a bet on DAC that is above both the projected future relative contribution of carbon removals to gross emissions abatement (20%, cf Prütz et al [50]) and what is agreed to be a more sensible ratio (10%, cf Geden and Schenuit [81]).We do not define bottlenecks for land use and water consumption as these arise only regionally and their spatially explicit assessment is beyond the scope of our analysis.Furthermore, we do not consider a net greenhouse gas emissions bottleneck because DACCS and DACCU lead to fundamentally different products: one leads to negative emissions that partially offset emissions from continued fossil fuel use in the short term; the other substitutes for emissionintensive fossil fuels and thus acts as a mitigation technology.
Finally, we evaluate and compare the reliance on existing or need for new transport infrastructure, its maturity, and the capital investments needed (cf supplementary methods 1.2.3).We define short-term bottlenecks if the initial deployment of DAC relies on new, not yet existing transport infrastructure, if the TRL of transport infrastructure is below 5, and if the capital investment in infrastructure by 2030 overshoots current investments midstream and downstream oil and gas infrastructure, including pipelines (€230 billion in 2021 [82]).
For the immaterial factors, we evaluate markets based on their accessibility for DAC-based products and services (negative emissions [Direct] or fuels/chemicals [Spillover]), their size and scalability, and the competition on these markets (cf supplementary methods 1.2.4).Unlike for capital investment needs (under the 'resource use' factor), we did not explicitly calculate the relevant values for our projected exponential deployment of DAC but reported them from the literature and market analysis websites.We define short-term bottlenecks if there are no existing markets or if future markets are foreseeably smaller than 1 GtCO 2 /year, where DAC-based products would be outcompeted by cheaper but less scalable products.We do not assess short-term bottlenecks for the competition criteria since most entrant green technologies are initially uncompetitive.
We assess regulations and international agreements based on the need for new regulations or international agreements and their conformity with existing frameworks.We define bottlenecks if the initial deployment of DAC requires new regulations and international governance arrangements before deployment can begin, or if the pathway is impeded by current arrangements.
Finally, for the social and behavioral factors, we evaluate and compare the conformity with current user practices and the polarization of the public debate (i.e. the balance between perceived risks and benefits).We define short-term bottlenecks as the adoption of disruptive user practices and perceived risks strongly outweighing perceived benefits.
Table 1.Analytical framework underpinning the comparison of the Direct and the Spillover pathway.For each pathway, we assess technical and immaterial factors, namely technology maturity, resources use, transport infrastructure, markets, regulation, international governance, and social and behavioral factors.For each element of this factor, we formalize evaluation criteria to assess the short-term feasibility of the pathway and the relative advantages of the pathways.In the 'Relative performance' column, we present the evaluation rationale for determining one pathway's advantage over the other across all evaluation criteria.

Technical factors
In terms of technical factors, the Direct pathway holds clear advantages in resource use and transport infrastructure, while both pathways perform similarly in terms of technological maturity.

Technological maturity
Neither pathway faces short-term technological maturity bottlenecks, as each process step has at least one technological option that has reached a demonstration level of maturity (minimum chain TRL > 5).The maturity level of the least mature irreplaceable component-monitoring technology for Direct pathway, and CO 2 reduction via reverse water-gas shift for Spillover pathway-is the same for both pathways (TRL 7). Figure 2 shows the TRL of each process step and of the respective technologies, including alternative technology options where applicable.
Restricting the Direct pathway to mature storage types such as saline aquifers and depleted oil and gas reservoirs, which are routinely used in carbon capture and storage and enhanced oil recovery projects [83], increases the maturity of the whole process chain [78,84].Some regions without access to mature storage types must instead rely on less mature storage technologies, such as in reactive rocks for in-situ mineralization (TRL 5) or supercritical injection in basalt rocks (TRL 3) [78].These storage methods have a lower risk of leakage as the CO 2 is transformed in carbonate rocks, but require site-specific exploration and risk and environmental impact assessment as they can contaminate the groundwater with mineral-rich plumes [85].As a result, the lead time from project conception to CO 2 injection at new sites can be three to ten years [86], creating a potential regional shortterm bottleneck [83,85].
In the Spillover pathway, using only the most mature electrolysis and CO 2 reduction technologies, an overall process with lowest TRL 7 can be achieved.However, some of the (still) least mature technology alternatives (e.g.anion exchange membrane and seawater electrolysis) hold the promise of future benefits such as increased cost effectiveness and energy efficiency [78].

Resource use
While both routes lead to an installed DAC capacity of 1 GtCO 2 , in terms of non-DAC components the Direct pathway requires 5-50 times less energy, 3-10 times less investment, and over 100 times less land than the Spillover pathway.In addition, the Direct pathway has 40%-200% lower greenhouse gas emissions, taking a net life-cycle emissions approach and assuming average grid emissions.In fact, the Spillover pathway uses substantially more resources to first produce green hydrogen, which then reacts with the air-captured CO 2 to synthetize fuels and chemical intermediates.The only resource that is used more in the Direct pathway is water, which is only required for in situ mineralization, where saline water can be used to minimize the depletion of freshwater resources.
Although the Spillover pathway uses significantly more resources than the Direct pathway, it does not meet our criteria for short-term bottlenecks, as its 2030 energy use and capital investment requirements are below 20% of current installed renewable energy (8000 TWh) and of current global capital investments in renewables (€500 billion).In addition, if the production of DAC-based hydrocarbons were entirely fueled by wind energy and low-temperature heat, the resulting net greenhouse gas emissions could be even lower than in the Direct pathway due to the substitution of fossil fuels with high well-to-tank emissions.Table 2 shows the resource use of the two pathways for both 2030 and 2050.

Transport infrastructure
The Direct pathway has the advantage of requiring only CO 2 transport infrastructure, while the Spillover also possibly relies on costly H 2 transport infrastructure.However, both pathways can similarly minimize their short-term dependence on new transport infrastructure by favoring integrated designs and reduce costs by co-developing infrastructure together with other industries.The maturity of both transport infrastructures is in a similar range.Table 3 summarizes the infrastructure needs of the two pathways, and the maturity, costs, and synergies of their transport infrastructure (pipelines).
In the Direct pathway, transport needs are minimal if DAC is co-located with storage sites.However, pipelines to transport CO 2 from the DAC plant to the storage site may help leveraging optimal location for DAC, such as where waste heat and cheap renewable energy are available [9,35,38,88], or where climatic conditions are ideal [88,89].In the long term, CO 2 pipelines will likely materialize independently of DACCS to transport CO 2 from point-sources ).The high cost of building CO 2 transport pipelines is therefore likely to be borne not only by the DACCS industry, but also by other hard-to-abate sectors with point-source capture.
In the short term, DAC-based hydrocarbons are likely to be produced in plants that integrate both DAC and hydrocarbon production to maximize energy efficiency [47].However, the large-scale hydrogen production needed in the Spillover pathway requires locations with abundant renewable energy potential, which may not overlap with ideal DAC sites (e.g.remote desert areas vs. humid areas with sources of waste heat) or with locations where demand for DAC-based hydrocarbons is high.This leads to tradeoffs between the transport of CO 2 , hydrogen, and final outputs.Since pipelines transporting hydrogen have lower mass flows than CO 2 [92], more pipelines are needed to transport hydrogen in the Spillover pathway, resulting in 2-6 times higher capital costs.However, the rapid expansion of the hydrogen sector is likely to require these infrastructures independently of DACCU [93].

Immaterial factors
In the short term, the Spillover pathway shows some advantages for most immaterial factors due to its partial alignment with the incumbent fossil fuel regime.In addition, the Direct pathway faces shortterm regulatory barriers that may hinder the initial deployment of DACCS in some regions.

Markets
The Direct pathway has certain disadvantages over the Spillover one since the product of DACCS-carbon removal certificates-are currently only traded on voluntary carbon removal markets, which are very limited in size and currently dominated by cheaper biochar-based certificates [31,94].While these markets are expected to grow to 60-750 MtCO 2 by 2040 [95], without further policy support, the demand for DACCS will unlikely grow at the pace needed, as only few buyers are willing to pay for the current price of DACCS.Policy-driven expansions and modifications of compliance markets are therefore needed for the Direct pathway to access larger market [30,31,96], as existing emission trading schemes do not recognize DACCS certificates and the prevailing emission prices do not serve as adequate incentives [97].Both access to and competitiveness of DACCS relative to other carbon removal methods on compliance carbon markets depend fully on public policies.Alternatively, policies could create dedicated carbon removal markets via instruments such as carbon take-back obligations, although this alone would not eliminate competition with other carbon removal methods [98][99][100][101].
On the other hand, the Spillover pathway can tap into existing large markets for fuels and chemical intermediates [102,103].Currently, DAC-based products can easily be certified for these markets, but they lack competitiveness due to their higher production costs compared to similar products from cheaper CO 2 feedstocks (e.g.fossil or biogenic CO 2 captured at point-sources).While policy support will be essential for market introduction, it may possibly be phased out in the long run.Emission trading system caps reaching net zero allowances would in fact force cheaper fossil-based products to exit the market.Until then, however, the effect of climate policies on the price of fossil-based products is uncertain: risky investments in fossil fuels and rising carbon prices could increase the price of fossil-based products [104].On the other hand, the electrification of some sectors as well as the diffusion of bio-based fuels, chemicals and materials reduce demand for fossil fuels, which could lead to plummeting prices.As a result, in these sectors, DAC-based products will not only have to close a wide price gap compared to fossil products but also to compete with alternative mitigation options.Nevertheless, DAC-based fuels and chemicals may find their niche in aviation, shipping, and industry, where electrification and biobased products are not scalable solutions [105,106].
Figure 3 summarizes the production cost of DACCS and DACCU, their potential markets, their respective size (scalability) and prices of incumbent products and services (competitiveness).

Regulations
While both pathways require new regulations that only partially overlap with current frameworks, the Spillover has an initial advantage since the Direct pathway faces additional short-term deployment bottlenecks.Underground CO 2 storage does not conform with some existing national regulations.However, amending these regulations and adopting new frameworks for CO 2 underground storage is feasible in the medium term, as has already been done in countries currently pursuing point-source carbon capture and storage [107,108].Moreover, liability and governance for long-term CO 2 storage need to be well-regulated before projects can start, as this determines the investment risk [109].The Spillover pathway, in contrast, can work immediately within existing regimes, as some DAC-based hydrocarbons are already approved for usage in a maximal blending of 50% vol with fossil fuels.Rather, this pathway needs new regulations to incentivize and accelerate the deployment of DACCU (e.g.blending quotas) [52,110,111] and to ensure that the hydrogen scaleup is sustainable.
Table 4 shows and describes the different pieces of legislation that are needed for each pathway, gives some examples of existing or planned legislation, and assesses the need for new regulations as well as the compliance of the pathways with existing legislation.

International agreements
While both pathways are similar in the short term because they can initially rely on bilateral agreements, the Spillover pathway may have an advantage over the Direct pathway in the long term.Indeed, largescale DACCS deployment is likely to require new or amended international agreements for either CO 2 transport or international trading of CO 2 removal certificates.Agreeing on this has proved to be politically difficult, as it does not fully conform with current arrangements and involves clarifying contentious details.Conversely, the Spillover pathway requires only transport agreements and bilateral trade, which are largely consistent with existing agreements for incumbent fossil fuels and chemicals.As shown in table 5, in the Direct pathway, international agreements are needed to allow countries without geological storage capacity to capture CO 2 domestically and transport the CO 2 to foreign storage sites [108,112].Countries can currently rely on bilateral agreements to circumvent current governance barriers to CO 2 transport, but amendments to existing multilateral agreements or new agreements are needed to facilitate the transport [112].Alternatively, countries without DACCS capacity can access CO 2 removal projects abroad through bilateral partnerships to establish DACCS hubs or bilateral 'internationally transferred mitigation outcomes' .Importantly, removal certificates can be traded internationally via multilateral mechanisms such as Article 6.4 of the Paris Agreement.However, these agreements need to be amended to explicitly include DACCS-specific methodological work, baseline setting, and monitoring and liability provisions, which can be a time-consuming process [30,113,114].
In the Spillover pathway, the required international governance changes are less challenging, as the pathway only requires bilateral trade and transport agreements for CO 2 , hydrogen, and for DACbased fuels and products.Although the Spillover pathway faces the same international governance needs and barriers to cross-border CO 2 transport as the Direct pathway, it is less dependent on CO 2 transport because it is not constrained by spatial limitations of geological storage sites.The international trade and transport of hydrogen needed to match hydrogen supply and demand [115] requires agreements that are consistent with the existing international governance architecture and do not create short-term bottlenecks.Similarly, agreements for the trade and transport of DAC-based hydrocarbons conform to international trade agreements of the incumbent fossil regime and therefore do not represent a bottleneck.

Social and behavioral factors
The Spillover pathway shows some advantages over the Direct pathway mainly because DAC-based products align well with existing user practices and CO 2 utilization tends to enjoy greater public acceptance than CO 2 storage [116].Table 6 shows the differences in alignment with current user practices, familiarity and perceived risks and benefits of the two pathways.
Unlike DAC-based hydrocarbons, DACCS does not replace an incumbent technology and therefore requires the adoption of new user practices (i.e.counterbalancing emissions with negative emissions).These practices are similar to current widespread offsetting practices, which are widely considered to be problematic [117].DACCS has also historically been associated with various concerns about its risks, such as its large energy requirements [118,119], its safety [8,85,120], and its entanglement with the fossil fuel industry and perverse incentives to continue polluting [121][122][123], which reduce public acceptance of DACCS [6,120,124].The benefits of DACCS, on the other hand, lie in its potential to not only enable net-zero greenhouse gas emissions in hard-to-abate sectors, but also to contribute to net-negative CO 2 emissions, although such impacts become climate-relevant only if absolute emissions fall below the rate of carbon removal [2,3,[5][6][7].This has led to a polarized debate characterized by enthusiasm on one side and rejection on the other side [61,62].Similarly, the benefits associated with DAC-based hydrocarbons, such as their potential to replace fossil fuels and chemicals and greatly reduce their climate impact [66,68,125,126] and to provide energy storage and flexibility [127], are balanced by some of their risks [128][129][130].Indeed, the social acceptability

Discussion
Our study shows that both pathways enabling DAC deployment face significant challenges and that neither has advantages across all factors (figure 4).
The Direct pathway offers advantages in terms of technical factors, particularly with respect to resources use and environmental impacts, while the Spillover pathway offers some short-term advantages in terms of immaterial factors, particularly markets and regulations.The disadvantages in the Spillover pathway are mainly related to the hydrogen production needed for DAC-based hydrocarbon production, while the Direct pathway faces significant regulatory barriers that need to be solved before initial deployment can take place, at least in some regional contexts.The barriers are thus fundamentally different, both in their nature and timing: in some regional contexts, the Direct pathway faces immediate barriers that make the pathway unviable until these barriers are removed, which may be a politically lengthy process.However, once regulatory barriers are overcome, the Direct pathway has lower costs, resource use, and infrastructure requirements than the Spillover pathway, highlighting its long-term advantages.Conversely, the Spillover pathway can start immediately, but will face increasing problems over time as resource requirements and costs increase with scale.The barriers to hydrogen production place the Spillover pathway on an unequal footing with the Direct pathway, as the latter does not produce fuels.Because hydrocarbons are likely to be needed in a climate-neutral future, the Direct pathway would de facto be equivalent to an 'emit and offset' pathway for the uses that are satisfied by the DAC-based hydrocarbons of the Spillover pathway.While previous studies have found energy and cost advantages of the 'emit and offset' over a pathway of absolute emissions abatement via DACCU in the aviation and chemical industries [66,67], the 'emit and offset' strategy may not be beneficial in sectors where there are cheaper options than DACCS to reduce emissions, e.g.where electrification is viable [1,136,137].
To kick-start the Direct pathway, its barriers need to be addressed, such as including and incentivizing DACCS in compliance markets or creating ad hoc carbon markets, adopting a comprehensive DACCS regulatory framework and international agreements, and increasing social acceptance.This may prove difficult in the short term, given the lack of markets and institutions and the polarized public and scientific debate on the deployment of DACCS.However, if alternative carbon removal methods, such as bioenergy with carbon capture and storage, become widely adopted in the future, they will also require CO 2 transport infrastructure, amendments to markets and regulations relevant to DACCS, and the inclusion of carbon removals in international agreements [74].The diffusion of these carbon removal methods would lay the groundwork for the deployment of DACCS in the Direct pathway and allow DACCS-supporting policies to leverage existing institutions rather than create new ones.
Taken together, these considerations suggest that policymakers would be well advised to pursue both approaches simultaneously: the DACCU approach to rapidly catalyze DAC deployment and learning, and embark on a path towards carbon-neutral synthetic hydrocarbons; and, at a later stage, after short-term bottlenecks have been resolved, the DACCS approach to enable scaled DAC deployment with lower resource and capital inputs, and to pave the way for negative emissions in the medium and long term.
Whether the Direct or Spillover pathway will be more successful in reality, however, depends on which set of factors-technical or immaterial-prove to be the larger obstacle.Our study does not answer this question, as this hinges on stakeholders' perception of and capacity to tackle these obstacles, which may vary in different regional and national contexts (e.g.how fast legislations can adapt to enable underground storage or market access).Moreover, the two sets of factors are not independent of each other, but are dynamically linked [75].Technical factors such as capital and energy intensity are influenced by deployment (e.g.efficiency improvements and cost reductions), which in turn is influenced by the policy support a technology receives [18].Policy support, conversely, is often determined by expectations of future costs and technological needs.A prime example is the competition between PV and concentrating solar power (CSP) [17,138].Thirty years ago, CSP was both cheaper and capable of producing dispatchable power through its integrated thermal storage capacity [139].Nevertheless, the small-scale nature of solar PV provided social benefits and led to policy support, while the modular nature of the technology allowed very high learning rates [140].This led to rapidly falling PV costs that soon outpaced CSP cost reductions, increasing the legitimacy of PV and decreasing that of CSP, eventually leading to the withdrawal of virtually all support for CSP [138,141].If such patterns were to be repeated in the case of DAC deployment, it would suggest that immaterial factors ultimately play a greater role; this would favor the shortterm pursuing of the Spillover pathway.However, the relative importance of factors may vary in different national and local contexts.For example, while generally a disadvantage, the high energy intensity of the Spillover pathway would not constitute a barrier in countries with abundant renewable energy resources.
By comparing two pathways to enable DAC in the near term across multiple dimensions for sociotechnical transitions, rather than just comparing technology performance, our study provides a comprehensive understanding of the different pathways and their implications.We show that both pathways are likely to be necessary and should be sequenced accordingly to leverage their respective advantages: the Spillover to kick-start DAC deployment and grow the technology, and the Direct to enable scaled DAC deployment at lower cost and using fewer resources.

Figure 1 .
Figure 1.Conversion route from air-captured CO2 to DAC-based end products and services under the two pathways considered, the Direct and the Spillover pathways.Dashed arrows show process steps that can be avoided depending on the process design.

Figure 2 .
Figure 2. Comparison of the maturity of each process technology alternative for the Direct and Spillover pathways.Maturity is assessed in terms of technological readiness level (TRL) both at aggregated process step level, and at the level of the alternative technologies available for each step.In red, we highlight those technologies with maturity below demonstration stage (6).

Figure 3 .
Figure 3.Comparison of market-relevant factors for the Direct and Spillover pathway, including available markets, their sizes, and competitivity 2022 and in the long-term.Values are derived from a review of the literature and of market analysis websites as outlined in the supplementary methods 1.2.4 and discussed in detail in the supplementary results 2.2.

Figure 4 .
Figure 4. Comparison of the Direct and Spillover pathways over all evaluation factors included in this analysis: technological maturity, resource use, transport infrastructure, markets, regulations, international agreements, and social and behavioral factors.Grey-colored cells show potential short-term bottlenecks to the deployment of DAC, while colored cells show whether a pathway has an advantage (green), disadvantage (red), or similar issues (yellow) relative to the other pathway.

Table 2 .
Comparison of resource use for the Direct and Spillover pathway in both 2030 and 2050 (for a detailed breakdown of resource use for different technology configurations in the Spillover pathway, cf supplementary results 2.1).The upper part of the table shows the input-output balance, the second part the environmental impacts (net greenhouse gas emissions, water, land use for renewable energy production), and the lower part the energy and capital use (hydrogen, electricity, heat, and capital investment).Values are reported in absolute units for the exponential DAC-uptake pathways and were calculated as detailed in supplementary methods 1.2.2.Values exclude the DAC component itself, as this is identical in both pathways and thus does not affect their comparison.Excluding spacing between wind turbines.Including spacing between turbines leads to hundred times larger land use, although turbines can be placed on land that has other uses (e.g.agricultural land).
a To be further processed into synthetic fuels and chemical intermediates.b Ground-mounted PV. c

Table 3 .
Comparison of transport infrastructure for the Direct and Spillover pathways.Evaluation criteria are the short-term reliance on transport infrastructure, its technological readiness level, the levelized cost of transport and capital investment, and synergies with other industries.

Table 4 .
Comparison of regulatory and legal frameworks under the Direct and Spillover pathway under the criteria of necessity of new regulatory frameworks and conformity with existing regulatory frameworks.For a detailed discussion of these results, cf supplementary results 2.2.

Table 5 .
Comparison of international governance under the Direct and Spillover pathways under the criteria of necessity of new governance agreements and conformity with current governance arrangements.For a detailed discussion of these results, cf supplementary results 2.3.

Table 6 .
Comparison of social and behavioral factors i.e. alignment with current user practices, familiarity, perceived risks and benefits, and legitimacy, for the Direct and the Spillover pathways.For a detailed discussion of these results, cf supplementary results 2.4.